Vibration control of 2-mass system using a neural network torsional torque estimator

A new control scheme using a torsional torque estimator based on a neural network is proposed and investigated for improving control characteristics of the high-performance motion control system. This control method presents better performance in the corresponding speed vibration response, compared with the disturbance observer-based control method. This result comes from the fact that the proposed neural network estimator keeps the self-learning capability, whereas the disturbance observer based torque estimator with the low pass filter should adjust the time constant of the adopted filter according to the natural resonance frequency determined by considering the system parameters varied. The simulation results show the validity of the proposed control scheme.